brand logo

Am Fam Physician. 1999;60(9):2613-2620

Epidemiologic and interventional studies have led to lower treatment targets for type 2 diabetes (formerly known as non–insulin-dependent diabetes), including a glycosylated hemoglobin level of 7 percent or less and a before-meal blood glucose level of 80 to 120 mg per dL (4.4 to 6.7 mmol per L). New oral medications make these targets easier to achieve, especially in patients with recently diagnosed diabetes. Acarbose, metformin, miglitol, pioglitazone, rosiglitazone and troglitazone help the patient's own insulin control glucose levels and allow early treatment with little risk of hypoglycemia. Two new long-acting sulfonylureas (glimepiride and extended-release glipizide) and a short-acting sulfonylurea-like agent (repaglinide) simply and reliably augment the patient's insulin supply. Combinations of agents have additive therapeutic effects and can restore glucose control when a single agent is no longer successful. Oral therapy for early type 2 diabetes can be relatively inexpensive, and evidence of its cost-effectiveness is accumulating.

The management of type 2 diabetes (formerly known as non–insulin-dependent diabetes) has changed greatly in recent years. Clear targets for glycemic control Clear targets for glycemic control have been established, the course of diabetes is better understood and new therapeutic agents have been introduced. Consequently, improved outcomes are now possible, even though establishing simple treatment guidelines remains a challenge.

This article focuses on the control of glucose levels with oral agents. Although factors such as lifestyle issues, lipid disorders, hypertension, smoking cessation and insulin therapy (used when oral agents alone are no longer successful) are highly important in patients with type 2 diabetes, they are beyond the scope of this discussion.

Targets for Control

The American Diabetes Association recommends a glycosylated hemoglobin (HbA1c) level of 7 percent or less as the target for glycemic control, with a level persistently over 8 percent serving as a signal to reassess and revise treatment.1 This target level is justified by results of large interventional trials showing the effectiveness of intensive treatment in reducing the ocular, nervous system and renal complications of diabetes.2,3

An HbA1c level of 7 percent corresponds to a before-meal capillary blood glucose level of 80 to 120 mg per L (4.4 to 6.7 mmol per L). Control this tight may be hard to achieve in patients with type 1 diabetes (formerly known as insulin-dependent diabetes) or type 2 diabetes of long duration. However, the new medications make this degree of control achievable in most patients with early type 2 diabetes.

Type 2 diabetes is progressive, and all treatments show secondary failure over time.4,5 Specifically, the HbA1c level creeps upward at 0.2 to 0.3 percent per year.4 Thus, in the typical patient, an HbA1c concentration of 7 percent will increase to 8 or 8.5 percent over five years unless treatment is escalated appropriately. Therefore, initial treatment must always be subject to revision.

Treatment of type 2 diabetes begins with efforts to improve lifestyle factors, including diet and exercise, but experience shows that glucose levels usually do not respond adequately to these measures even in the short run. In the United Kingdom Prospective Diabetes Study,6 only 15 percent of patients with newly diagnosed diabetes who were advised to make dietary changes and increase their level of exercise achieved the glycemic target. With longer treatment using only diet and exercise, glycemic control progressively worsened. Pharmacotherapy is initiated when dietary measures and exercise prove ineffective.

Agents for the Management of Type 2 Diabetes

The agents used to manage type 2 diabetes can be divided into two groups: those that augment the patient's supply of insulin and those that enhance the effectiveness of insulin.7


Other than injected insulin, two kinds of agents augment the body's insulin supply (Table 1).7 The sulfonylureas and the meglitinides increase the secretion of endogenous insulin, as long as pancreatic beta-cell function remains. All of the insulin-augmenting agents act by binding to parts of a receptor complex on the surface of the beta cell.

The rightsholder did not grant rights to reproduce this item in electronic media. For the missing item, see the original print version of this publication.

Insulin-augmenting agents are ineffective in patients with juvenile-onset type 1 diabetes. They are also poorly effective in adults who are developing type 1 diabetes.

Two new long-acting sulfonylureas are now available: glimepiride (Amaryl)8,9 and extended-release glipizide (Glucotrol XL).10,11 The first meglitinide to become available is repaglinide (Prandin).12,13

The various insulin-augmenting agents have equivalent therapeutic power but differ in duration of action and site of clearance. Repaglinide and tolbutamide (Orinase) are the most rapid- and short-acting agents, whereas chlorpropamide (Diabinase), extended-release glipizide and glimepiride are the slowest and longest acting agents.

At present, glyburide (Micronase, but mainly the generic form), extended-release glipizide and glimepiride are the oral antidiabetic agents most widely used in the United States. Glyburide is inexpensive; however, for full effectiveness, it must be taken twice daily, and it has an active metabolite that accumulates when renal function declines. Extended-release glipizide and glimepiride are taken once daily, and their clearance depends very little on renal excretion.


The insulin-assisting agents (Table 2)7 include the following: metformin (Glucophage), which is a biguanide14,15; acarbose (Precose)16,17 and miglitol (Glyset),18,19 which are α-glucosidase inhibitors16,17; and pioglitazone (Actos), rosiglitazone (Avandia) and troglitazone (Rezulin), which are thiazolidinediones.20,21 The mechanisms of the classes of agents differ widely.

The rightsholder did not grant rights to reproduce this item in electronic media. For the missing item, see the original print version of this publication.

Metformin acts mainly on the liver. It improves the response of this organ to insulin and reduces overnight glucose production and fasting hyperglycemia. At higher dosages, metformin may reduce food intake and help with weight control.

Acarbose and miglitol delay the digestion and absorption of complex carbohydrates. Although these agents do not increase the response to insulin in any tissues, their ability to limit glucose increases after meals can reduce plasma insulin levels and the need for injected insulin.

Pioglitazone, rosiglitazone and troglitazone improve the response of muscle and adipose tissue to insulin in patients who are otherwise relatively unresponsive, notably the extremely obese.

The side effects and limitations of the insulin-assisting agents also differ. Metformin is not metabolized and must be excreted by the kidney. Because high blood levels of metformin can cause fatal lactic acidosis, this oral agent cannot be used when the serum creatinine concentration exceeds 1.4 mg per dL (120 μmol per L) in women or 1.5 mg per dL (130 μmol per L) in men. When first taken, metformin often causes nausea or diarrhea. These effects persist in 5 percent of patients taking a low dosage of metformin and in 20 to 25 percent of patients treated with a full dosage.22

Acarbose and miglitol are quite safe, but they often cause flatulence, especially when the dosage is increased rapidly. Neither agent should be used in patients with significant intestinal disorders.

Troglitazone sometimes causes fluid retention and weight gain. Rarely, it causes fatal liver injury. Because of the threat of liver toxicity, serum alanine aminotransferase (ALT) measurements should be obtained frequently (i.e., before treatment, monthly for 12 months and every four months thereafter). Troglitazone should be stopped immediately if serum ALT levels become elevated to three times the upper limit of normal.

Whether the two new thiazolidinediones (rosiglitazone and pioglitazone) will be entirely free of liver toxicity remains to be seen. To date, no cases of drug-associated liver failure have been verified with these agents, and less frequent ALT measurements (every two months for the first year) are advised.


Dose-response relationships and therapeutic effects have been defined for all oral agents when they are used singly in unselected populations of patients (Tables 1, 2 and 3). For most agents, the therapeutic effect is achieved with a dosage below the maximum allowed dosage.

Class or generic nameBrand nameEffect on glucose levelAverage reduction of HbA1c level (%)Patients best suited for treatment
AcarbosePrecoseDecreases postprandial increase0.5 to 1Patients with high postprandial glucose levels
MetforminGlucophageDecreases fasting and 24-hour mean levels1 to 2Obese patients with recently diagnosed type 2 diabetes
RepaglinidePrandinDecreases fasting and postprandial levels1 to 2Patients with recently diagnosed type 2 diabetes who have high postprandial glucose levels
Sulfonylureas*Decrease fasting and 24-hour mean levels1 to 2Patients with recently diagnosed type 2 diabetes
TroglitazoneRezulinDecreases fasting and postprandial levels0.5 to 1.5Patients who are obese or insulin-resistant

The sulfonylureas and repaglinide typically reduce HbA1c levels by 1 to 2 percent. With glyburide, most of the therapeutic effect is achieved with a dosage of 5 mg twice daily. Little is gained by increasing the dosage of extended-release glipizide above 5 mg per day10 or the dosage of glimepiride above 4 mg per day.8

The full effect of metformin is the same as that of the sulfonylureas. A recent U.S. trial showed that the maximal effect of metformin is achieved at a dosage of 2,000 mg per day, with a surprisingly strong effect occurring at a dosage of 500 mg taken once or twice daily.14

In unselected populations, acarbose and troglitazone have been found to reduce HbA1c levels by 0.5 to 1 percent16,17 and 0.5 to 1.5 percent,20,21 respectively. Early reports of similar studies of miglitol, pioglitazone and rosiglitazone have shown similar effects. These findings suggest that the α-glucosidase inhibitors and thiazolidinediones are generally less powerful than the sulfonylureas or metformin, although this limitation can be attributed partly to the wide variability among individual patients.

The maximum daily dosage of acarbose or miglitol is 100 mg taken three times daily, but nearly full effects occur when these agents are given in a dosage of 50 mg three times daily with meals. The glycemic effect of troglitazone is often greater at a dosage of 600 mg per day than at a dosage of 400 mg per day. Unpublished reports suggest that the maximal effect of pioglitazone occurs at a dosage of 45 mg per day, and the maximal effect of rosiglitazone occurs at a dosage of 4 mg twice daily.

The differences in consistency of benefit for various agents may affect clinical decisions. Most patients with recently diagnosed type 2 diabetes respond well to sulfonylureas, repaglinide or metformin. In contrast, acarbose and troglitazone are more effective in some patients than in others. Acarbose works best in patients who have prominent post-prandial hyperglycemia and are less disturbed by flatulence. Troglitazone works best in patients with marked insulin resistance, notably the extremely obese, some of whom may respond quite dramatically to this agent.

Initiation of Treatment

Universal agreement is lacking on the level of glycemic control at which pharmacologic therapy needs to be initiated. Ideally, diabetes is diagnosed early, when the fasting plasma glucose level is between 126 and 140 mg per dL (7.0 to 7.8 mmol per L) and the HbA1c concentration is in the range of 6.5 to 7.5 percent.5,7 At these levels, agents that augment blood levels of insulin can sometimes cause hypoglycemia, whereas insulin-assisting agents used alone rarely cause this problem.

Metformin can be quite effective in a dosage of 500 mg taken once daily before a major meal (to maximize its satiety effect) or at bedtime (to control fasting hyperglycemia). The dosage can be titrated to 850 mg once daily, 500 mg twice daily, or higher. Patients should be instructed to stop taking metformin if they develop an illness that causes dehydration or otherwise threatens renal clearance of this agent.

Acarbose or miglitol is a better initial choice in patients who have renal impairment and thus cannot use metformin, especially if their fasting glucose level is below 140 mg per dL but their HbA1c concentration is well above 7.5 percent (suggesting marked postprandial hyperglycemia). Some older patients show this pattern.

Acarbose and miglitol are best started at a dosage of 25 mg taken once daily with a meal for two weeks. Then the dosage is increased to 25 mg taken twice daily at meals for two more weeks. Finally, the dosage is increased to 25 mg taken three times daily at meals. If necessary, the dosage may be increased to 50 mg with each meal. Gradual titration induces starch-splitting enzymes in the distal small intestine. This limits the passage of carbohydrate into the colon and thus reduces flatulence.

The role of thiazolidinediones as initial therapy is incompletely defined.23 Because of concerns about liver toxicity, troglitazone is no longer recommended as monotherapy. However, the U.S. Food and Drug Administration has labeled pioglitazone and rosiglitazone for use as monotherapy.

Although early treatment is desirable, most patients are identified later in the course of diabetes, when the fasting glucose level is much higher than 140 mg per dL, and the HbA1c concentration is well above 7.5 percent.5,7 At these levels, sulfonylureas or repaglinide can be used safely. Repaglinide is new and still under evaluation. However, the sulfonylureas have been well tested and have attractive features, including fast and predictable effects on glucose levels, few side effects, once-daily dosage and low cost.

To minimize the risk of hypoglycemia, the starting dosage of a sulfonylurea should be low. For example, glyburide should be initiated in a dosage of 1.25 or 2.5 mg once or twice daily, and glimepiride should be started in a dosage of 1 mg once daily. The lowest available dosage of extended-release glipizide is 5 mg per day, which is usually the maximal effective dosage. The starting dosage of repaglinide is 1 mg taken three times daily with meals.

Metformin, acarbose, miglitol and, perhaps, some thiazolidinediones may also be used as initial therapy in patients with higher glucose levels.

No large studies objectively define which patients should be started on which agent, but clinical features may suggest a preferred agent. The need for low cost or a quick response favors a sulfonylurea. An ongoing struggle with weight control argues for the use of metformin. Acarbose has appeal for patients with mainly postprandial hyperglycemia. A thiazolidinedione may have a role in patients judged to be highly insulin resistant and for whom other agents are ineffective as monotherapy.

Combinations of Oral Agents

Because a single agent is unlikely to maintain glycemic control for more than a few years, combination therapy is becoming quite common. Fortunately, the presently available classes of agents have fully additive therapeutic effects but independent side effects. The benefits of combining sulfonylureas with metformin,15,24,25 sulfonylureas with troglitazone26 and sulfonylureas with acarbose17,27 are well documented.


The combination of a sulfonylurea with metformin has been most widely used.15,24,25 When a sulfonylurea alone or metformin alone fails, the other agent can be added in a gradually titrated dosage. For example, treatment with extended-release glipizide in a dosage of 5 mg per day can be enhanced by the addition of metformin in a dosage of 500 mg once daily. As necessary, the dosage of metformin can be titrated to 500 mg taken two or three times per day. Similarly, glimepiride in a dosage of 1 mg per day may be added to metformin in a nearly maximum dosage of 1,500, 1,700 or 2,000 mg per day, and the dosage of glimepiride can be titrated to 4 mg per day as needed.


The combination of a sulfonylurea plus a thiazolidinedione also seems destined to be widely used. In extremely obese patients with poor glycemic control achieved using a sulfonylurea alone, the HbA1c concentration may decrease by more than 1.5 percent after the addition of troglitazone.26 Some patients treated with this combination gain as much weight as they would on starting insulin therapy, but the clinical significance of this side effect is unknown. Troglitazone is best absorbed if it is taken with a meal. The starting dosage can be 200, 300 or 400 mg per day. The starting dosages of pioglitazone and rosiglitazone are 15 mg per day and 4 mg per day, respectively, and both agents can be taken with or without food. The clinical response to thiazolidinediones develops slowly, beginning in a week or two and continuing up to at least three months. Fasting glucose levels obtained one month after the initiation of therapy can be used to judge whether the patient is responding to therapy.


Whatever combination is used, some common principles apply. Combining drugs is usually more effective than stopping one agent and substituting another. Adding a second agent is usually better than increasing the dosage of an agent that is already being given in a nearly maximum dosage. Secondary failure of two-drug combinations should be expected eventually. In some patients, three-drug combinations may be useful, although experimental testing is lacking. In most patients, the failure of two oral agents used together calls for the use of insulin alone or in addition to oral agents.

Measurement of Blood Glucose and HbA1c Levels

The success of oral therapy must be assessed by capillary blood glucose tests performed by the patient and by HbA1c concentrations measured during office visits. HbA1c levels should be determined at least twice a year.28

Older glucose strips and meters give values about 15 percent lower than laboratory measurements of plasma or serum glucose levels, whereas newer strips are calibrated to match laboratory methods. The target range for the before-meal glucose level is 80 to 120 mg per dL (4.4 to 6.7 mmol per L).1

All patients who are receiving oral antidiabetic drug therapy should learn how to test their blood glucose levels, but the testing schedule depends on the individual patient and the treatment.28 In general, morning fasting tests are most valuable. However, tests performed before the noon and evening meals can also be helpful. Tests performed after meals may be most useful for monitoring the effects of acarbose or miglitol. The frequency of testing can vary from several times a week (for patients who are taking agents that do not cause hypoglycemia and whose glucose levels are under stable control) to several times a day (for patients who are beginning treatment with a sulfonylurea or repaglinide).

Dosages of sulfonylureas, repaglinide, α-glucosidase inhibitors and metformin may be adjusted as often as weekly. Dosages of thiazolidinediones can be adjusted every one to two months.

Cost Issues

No prospective studies have compared the costs and savings for optimal oral therapy versus no pharmacologic treatment. However, the costs of the agents themselves are known. For example, at average wholesale prices, once-daily treatment with 10 mg of extended-release glipizide or 4 mg of glimepiride costs about $20 per month ($240 per year). Oral therapy with two 500-mg metformin pills per day costs about $32 per month ($384 per year). Combination therapy with this dosage of metformin plus one of the sulfonylureas would cost $42 per month ($504 per year). Compared with careful lifestyle treatment alone, oral drug therapy for type 2 diabetes requires little increase in the frequency of glucose testing, HbA1c measurements or office visits.

Early treatment with a sulfonylurea can reduce HbA1c from a typical starting level of 9 percent to a level of 7 to 7.5 percent.9,10 When secondary failure occurs, adding metformin in a dosage of 500 mg twice daily can reduce the HbA1c level from 8 percent to about 7 percent.14 In most patients, a sulfonylurea followed by metformin should control recently diagnosed diabetes for up to five years.

One recent study suggested the social and economic benefits of such treatment.29 In this study, more than 500 patients with newly diagnosed type 2 diabetes took extended-release glipizide or placebo for three months, and psychosocial and behavioral outcomes were measured. HbA1c concentration declined from 9.3 to 7.5 percent in the treatment group. Many symptoms, general perceived health and cognitive function also improved in the treated patients. This improved sense of well-being was accompanied by fewer sick days, less absenteeism, more productivity, better employment histories and less use of health facilities, with resultant cost-savings. The improvement of glucose control should reduce long-term ocular and renal complications by at least 50 percent.2,3

Another cost-benefit analysis described much greater treatment costs than those presented in this article. However, the investigators in this study concluded that early intervention is justified in patients with type 2 diabetes.30

Continue Reading

More in AFP

More in Pubmed

Copyright © 1999 by the American Academy of Family Physicians.

This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP.  See permissions for copyright questions and/or permission requests.